DeepSeek
DeepSeek publishes open weights and prices well below the US labs while scoring near the top on coding and reasoning benchmarks. Data is processed in China, which rules it out for some compliance regimes.
- Models listed
- 4
- Currently offered
- 4
- Cheapest language model
- DeepSeek V4 Flash 0731 $0.1488 / 1M tokens
- Headquarters
- CN
- API status
- Operational
- Training policy not published
- Self-hostable
Models by category
Language models 4
| # | Model | Input | Output | Context | Released | Capabilities | Price change |
|---|---|---|---|---|---|---|---|
| 1 |
DeepSeek V4 Pro
DeepSeek
|
$1.32 | $3.96 | 1M | August 2026 | ReasoningTool calling | |
| 2 |
DeepSeek V4.1 Flash
DeepSeek
|
$0.30 | $1.20 | 1M | October 2026 | VisionReasoningTool calling | |
| 3 |
DeepSeek V4 Flash 0731
DeepSeek
|
$0.12 | $0.24 | 1.05M | Jul 31, 2026 | ReasoningTool calling | |
| 4 |
DeepSeek V4 Flash Vision Exp
Preview
DeepSeek
|
$0.22 | $0.65 | 1.05M | Aug 21, 2026 | VisionReasoningTool calling |
Frequently asked questions
How many AI models does DeepSeek offer?
This catalogue lists 4 models from DeepSeek, 4 of them currently offered, across 4ats.
Does DeepSeek train on API data?
Training policy not published.
What is the cheapest DeepSeek language model?
DeepSeek V4 Flash 0731, at $0.1488 / 1M tokens (blended, 3 input tokens to 1 output).